HAL Id: hal-00384745
https://hal.archives-ouvertes.fr/hal-00384745
Submitted on 15 May 2009
HAL is a multi-disciplinary open access
archive for the deposit and dissemination of
sci-entific research documents, whether they are
pub-lished or not. The documents may come from
teaching and research institutions in France or
abroad, or from public or private research centers.
L’archive ouverte pluridisciplinaire HAL, est
destinée au dépôt et à la diffusion de documents
scientifiques de niveau recherche, publiés ou non,
émanant des établissements d’enseignement et de
recherche français ou étrangers, des laboratoires
publics ou privés.
Thermocapillary manipulation of microfluidic droplets:
Theory and applications
François Gallaire, Charles Baroud, Jean-Pierre Delville
To cite this version:
François Gallaire, Charles Baroud, Jean-Pierre Delville. Thermocapillary manipulation of
microflu-idic droplets: Theory and applications. International Journal of Heat and Technology, International
Information and Engineering Technology Association, 2008, 26 (1), pp.161-166. �hal-00384745�
THERMOCAPILLARY MANIPULATION OF MICROFLUIDIC
DROPLETS: THEORY AND APPLICATIONS
Fran¸cois Gallaire
1, Charles N. Baroud
2, Jean-Pierre Delville
31
Laboratoire JA Dieudonn´e, Universit´e de Nice Sophia-Antipolis
2
LadHyX, Ecole Polytechnique, 91128 Palaiseau Cedex
3
CPMOH, Universt´e de Bordeaux I, 351 Cours de la lib´eration, F-33405 Talence Cedex
September 12, 2007
Abstract
It was recently demonstrated by our group that a focused laser beam could be used to produce a net force on a moving microfluidic drop. The aim of the paper is to establish a scaling law for this net force by a examining the closely related but simpler situation of a very thin stationary circular drop of fixed shape submitted to a thermocapillary (Marangoni) stress. This leads us to recall the depth-averaged model for a microfluidic pancake-like undeformable drop submitted to a thermocapillary forcing. Our numerical method to solve the associated equations is then introduced and validated. In the case of a localized heating and for an ‘inverse’ Marangoni effect (i.e. the surface tension increases with temperature) mimicking the experimental situation of a focused laser beam impinging on a surfactant laden water-oil interface, the flow field is computed and compared to experimental observations. The viscous shear stresses (normal and tangential) and the pressure force are then computed on the interface, yielding a simple expression for the total force acting on the droplet. Further numerical investigations are conducted and enable us to propose a scaling law for the net force combining all pertinent parameters.
1. INTRODUCTION
Effective actuation of microfluidic drops may appear as simple, but when both inertial and buoyancy effects are negligible, the manipula-tion of water in oil drops turns out to be diffi-cult to achieve. Furthermore, several more com-plicated operations are necessary if one wishes to use droplets as microreactors in microfluidic lab-on-a-chip systems. These operations include the ability to create a reaction (fusion of drops and mixing of a drop’s content), test the results (sort drops), or sample their contents by divid-ing them asymmetrically.
We have recently demonstrated that a local-ized heating from a laser source on the surface of the drop may be used to produce these op-erations through the thermocapillary effect, by which the surface tension of an interface varies
with the temperature. In our recent
experi-ments [1], we showed that a water in oil drop, transported by the external oil flow, may be blocked if submitted to a focused laser spot. This blocking may last a few seconds, imply-ing that the force due to the drag from the oil is
balanced (at least temporarily) by a force which is generated by the laser heating. This block-ing force was then combined with the geome-try of the micro-channel to succeed in most of the aforementioned operations (merging, sorting and dividing).
The effect of temperature gradients on drops has been studied since the 1950’s because of its applications in microgravity situations [2, 3] and has been recently revived by the applications in microfluidics [4, 5]. The generally accepted de-scription [2, 4] is that surface tension decreases with temperature. Therefore if a drop is placed in a temperature gradient, it will experience flow along the interface from the hot to the cold side. In a low Reynolds number situation, this flow very rapidly (immediately for Re → 0) produces a matching flow in the outer fluid from the hot to the cold region. Therefore by conservation of matter, the drop must “swim” towards the hot side.
This description is not compatible with our observations as deduced from figure 1 which de-picts a cross-shaped microchannel in which oil
Laser off
Laser on
Laser
(d)
(e)
(f)
(a)
(b)
(c)
Figure 1: The water interface is temporarily blocked when it crosses the path of the focused laser in response to a force directed away from the hot spot, resulting in the production of a larger drop.
flows from the top and bottom branches and wa-ter from the left branch. Wawa-ter drops in oil are produced and flow down the right hand branch in the absence of a laser. However, when the water interface crosses the laser spot, it is com-pletely blocked, indicating that there exists a force acting from the right to the left, i.e. coun-teracting the viscous drag produced by the ex-ternal flow. Unlike the usual Marangoni force, this force is directed from the heated region cen-tered on the focusing point of the laser to the cold regions of the flow away from the laser thereby balancing the drag from the external oil flow and blocking the droplet.
Rather than simply ruling out a thermocapil-lary origin of the observed force, we have intro-duced tiny toner particle in the fluids in order to visualize in detail the direction of the flow in our experiments. Figure 2(a) shows that the flow is directed form the cold regions towards the hot spotalong the interface, pointing to an ‘in-verse’ Marangoni effect. Without ignoring that, for pure liquids, the direction of Marangoni flow along the interface is directed from the hot (low surface tension) to the cold (high surface ten-sion) regions, this opposite finding is consistent with previous studies [6, 7] which have shown an increase of surface tension with temperature in the presence of surfactants.
2. GOVERNING EQUATIONS AND NUMERICAL METHOD
Our experiments were conducted in high as-pect ratio channels. The typical dimension in the transverse direction is about 500 µm, while the typical dimension in the thickness is about
laser
(a)
main flow
−2 −1.5 −1 −0.5 0 0.5 1 1.5 2 −2 −1.5 −1 −0.5 0 0.5 1 1.5 2 r θ x yFigure 2: (a) Overlay of 100 images from a video sequence, showing the motion of seeding parti-cles near the hot spot. Note that the motion along the interface is directed towards the hot spot. Channel width is 140 µm.(b) Streamfunc-tion contours obtained by solving the depth-averaged model described in the text. Red con-tours indicate clockwise flow and blue concon-tours
indicate counter-clockwise flow. h/R = 0.2,
w/R = 0.5.
20-30 µm. With typical velocities of 1mm/s, the Reynolds number is seen to be very small. For these two reasons, we use a Hele-Shaw-like approximation, based on the depth-averaged Stokes equations as Boos and Thess [8], Bush [9] or Nadim et al. [10]. Note that, contrary to the Hele-Shaw approximation, where the depth-averaged velocity field derives from a potential and cannot fulfill the Marangoni induced tan-gential shear stress discontinuity at the inter-face, higher-order terms are kept in the
equa-tions to yield a 4th-order biharmonic-type
er eθ h y x z R
Figure 3: Schematic of the depth-averaged
model.
Let us consider a circular drop of radius R in an infinite domain, as sketched in Fig 3. As-suming a parabolic profile in the small dimen-sion (z) and introducing a streamfunction for the mean velocities in the plane of the channel, the depth averaged equations derived from the Stokes equations valid in each fluid are
1 r ∂ ∂rr ∂ ∂r+ 1 r2 ∂2 ∂θ2 1 r ∂ ∂rr ∂ ∂r+ 1 r2 ∂2 ∂θ2 − 12 h2 ψ = 0, (1) where the depth-averaged velocities may be
retrieved from uθ = −∂ψ/∂r and ur =
1/r(∂ψ/∂θ). At the drop interface (r = R), two kinematic boundary conditions are imposed; zero normal velocity
ψ(1)= ψ(2)= 0, (2)
and the continuity of the tangential velocity
∂ψ(1)
∂r =
∂ψ(2)
∂r . (3)
The normal dynamic boundary condition is not imposed since the geometry of the drop is imposed and the transverse curvature of the meniscus is assumed to counterbalance the pres-sure difference at the interface. Finally, the tan-gential dynamic boundary condition which takes into account the Marangoni stress is
µ1r ∂ ∂r u1 θ r − µ2r ∂ ∂r u2 θ r = −γr′dTdθ, (4)
where µ1,2 are the dynamic viscosities and u1,2θ
are the azimuthal velocities in the drop and
the carrier fluid, respectively, and where γ′ =
∂γ/∂T is the surface tension to temperature gra-dient and T (r, θ) is the imposed temperature field.
A pseudo-spectral discretization technique combining a Tchebitscheff collocation method in
the radial direction with respectively Nr(1) and
Nr(2) collocation points and a Fourier expansion
along the azimut with Nθmodes is implemented
in each fluid together with the 4 boundary con-ditions at the interface. The symmetry condi-tions at the origin in the inner fluid are taken implicitly into account as in [11] and an outer
radius of rmax = 8R is chosen to impose the
far field boundary conditions in the outer fluid. The code, written in matlab, can be run on a laptop in a few minutes as long the memory re-quirements remain reasonable (i.e. as long as h/R is not too small, preventing the appear-anceof very thin boundary layers in each fluid). The convergence of the results with increasing resolution has been checked. We use in general
N1
r = 40, Nr2= 40, Nθ= 40.
As a typical test-case, the Marangoni induced flow around a drop in a linear temperature gra-dient and with a positive surface tension to
tem-perature gradient γ′ > 0 is considered and our
results are compared to the analytical solution of [8] in figure 4, displaying an excellent agree-ment, with errors less than 0.1% for h/R = 0.2.
T
∆
Figure 4: Superposition of the analytical solu-tion obtained by Boos and Thess [8] (full iso-lines) and our numerical computation (dashed isolines) in presence of a linear temperature
gra-dient. h/R = 0.2 and µ2 = µ1. The color
con-vention is similar to Fig. 2b.
3. LOCALIZED HEATING
Thess[8] is only valid for a linear temperature gradient and a numerical approach becomes compulsory when the heating is localized, so as to mimic the experiment. According to the ex-perimental observations, the surface tension to
temperature gradient γ′ is taken positive and
may be estimated at γ′ ∼ 1 mNm−1K−1 [6,
7]. For simplicity, we approximate the steady state temperature distribution using a Gaussian form corresponding to a hot region of size w,
T (x, y) = ∆T exp−(x−R)2 +y2w2 , where ∆T is the
maximum temperature difference between the hot spot and the far field and w corresponds to the hot spot size and is significantly larger
than ω0. The equations are subsequently made
nondimensional using ∆T as temperature scale,
R as length scale, Rγ′∆T as force scale and
µ1+µ2
γ′∆T as time scale, the remaining
nondimen-sional groups being the aspect ratio h/R, the nondimensional spot size w/R and the viscosity
ratio ¯µ2 = µ2/(µ1+ µ2). Nondimensional
vari-ables are denoted in the sequel by an overbar. A typical flow field is shown in Fig. 2(b). As in the experimental image, and despite the dif-ference in the drop shape and in the presence of lateral walls, four recirculation regions are clearly visible and the flow is seen to converge towards the hot region along the interface.
0 0.2 0.4 0.6 0.8 1 −0.3 −0.2 −0.1 0 w/R pressure stress total force h/R=0.05 h/R=0.1 h/R=0.2 tangential shear normal shear stress F
(c)
Figure 5: Nondimensional force ¯F plotted as a
function of w/R for various aspect ratios h/R
(individual points) for ¯µ2= 3/4. The lines
cor-respond the asymptotic expression Eq. 14. Inset shows the distribution along the azimuthal di-rection of the pressure, normal and tangential shear stresses for h/R = 0.2, w/R = 0.5. Their sum yields the total force.
Three stress terms apply a net force on the
drop: the pressure field as well as the viscous shear stress in the tangential and normal direc-tions. The resulting x-component of the total force on the drop is given by the projections of these three contributions, integrated along the azimuthal direction, as shown in the inset of Fig. 5 for h/R = 0.2 and w/R = 0.5, where these three integrands and their sum are repre-sented (once suitably normalized) as a function of the azimut by the signed distance of the cor-responding curve from the unit circle. Note that the global x-component of the force is negative (the y component vanishes by symmetry) and is thereby oriented away from the laser. It may be shown in addition that the integral of the wall friction vanishes since the drop is stationary.
Let us compute the pressure field as well
as the normal ¯σ¯r¯r = 2¯µ2∂∂¯u¯rr¯ and
tangen-tial ¯σrθ¯ = ¯µ2 1r¯∂∂θu¯r¯ +∂∂¯u¯rθ −u¯r¯θ viscous shear
stresses in the external flow on the interface at ¯
r = ¯R = 1.
Since ¯ur¯ vanishes on the interface at ¯r = 1
independently of θ, the viscous tangential shear force reduces to ¯ Ft= −¯µ2 h R Z 2π 0 ∂ ¯uθ ∂¯r − ¯ uθ ¯ R ¯ r=1 sin(θ) ¯Rdθ = ¯µ2 h R Z 2π 0 −∂ ¯uθ ∂¯r + ¯uθ ¯ r=1 sin(θ)dθ. (5)
The viscous normal force is reexpressed using
the continuity equation ∂u¯¯r
∂r¯ = − 1 ¯ r ∂¯uθ ∂θ − ¯ u¯r ¯ r and ∂u¯r¯ ∂¯r
r=1¯ = −1r¯∂¯∂θuθ on the drop interface,
¯ Fn= 2¯µ2 h R Z 2π 0 −1¯ R ∂ ¯uθ ∂θ ¯ r=1 cos(θ) ¯Rdθ, (6) and after an integration by part
Fn= −2¯µ2¯h
Z 2π
0
(¯uθ)r=1¯ sin(θ)dθ. (7)
The pressure force is given by ¯ Fp= h R Z 2π 0 − ¯p|r=1¯ cos(θ) ¯Rdθ (8)
An integration by parts first gives ¯ Fp= h R Z 2π 0 ∂ ¯p ∂θ ¯ r=1 sin(θ) ¯Rdθ (9)
The term ∂ ¯p/∂θ may be retrieved from the
equations ¯ µ2 1 ¯ r ∂ ∂¯r¯r ∂ ∂¯r− 1 ¯ r2+ 1 ¯ r2 ∂2 ∂θ2 − 12R2 h2 ¯ uθ + ¯µ2 2 ¯ r2 ∂ ¯ur¯ ∂θ = 1 ¯ r ∂ ¯p ∂θ. (10)
Since ¯ur¯vanishes on the interface at ¯r = 1
inde-pendently of θ, the term ¯r22
∂u¯(2)r¯
∂θ is zero at ¯r = 1
and the pressure force becomes ¯ Fp = ¯µ2h R Z 2π 0 ¯ R ∂ 2 ∂¯r2+ 1 ¯ R ∂ ∂¯r −¯1 R2 + 1 ¯ R2 ∂2 ∂θ2 − 12R2 h2 ¯ uθ sin(θ) ¯Rdθ ¯ r=1 . (11) A double integration by parts leads
Fp = ¯µ2 h R Z 2π 0 ∂2 ∂¯r2+ ∂ ∂¯r− 2 −12R 2 h2 ¯ uθ sin(θ)dθ ¯r=1 . (12) The final expression for the nondimensional force is obtained by adding (5), (7) and (12)
¯ F = ¯Ft+ ¯Fn+ ¯Fp = ¯µ2 h R Z 2π 0 ∂2 ∂¯r2 − 12R2 h2 − 3 ¯ uθsin(θ)dθ r=1¯ . (13) Results of the computations for different aspect ratios (h/R) and hot spot size (w/R) and for ¯
µ2 are shown in Fig. 5, where it appears that
the data is well fitted (see the solid lines) by an expression as simple as ¯ F ≃ −α¯µ2 h R w R, (14)
with the prefactor α ≃ 3.5. This expression may be understood by replacing the implicit (since it
involves ¯uθ which is deduced from the solution
of the biharmonic equation (1)-(2)) expression (13) by its explicit boundary layer approxima-tion (in the spirit of [8, 10]) in the limit where h is small compared to all characteristic in-plane length scales (h ≪ R, w) ¯ F ≃ −2¯µ2 h R Z 2π 0 ¯ T (¯r = 1, θ) cos(θ)dθ. (15)
This expression, which requires only the knowl-edge of the temperature profile, may be further approximated in the limit of small w/R by
¯
F ≃ −2√π ¯µ2h
R w
R, (16)
in fine agreement with the fitted value of α since
2√π ≃ 3.55. The resulting dimensional scaling
is easily retrieved
F ≃ −2
√
π∆T γ′hwµ2
R(µ1+ µ2) . (17)
With µ1 = 10−3 Nm−2s (water) and µ2 = 3µ1
(hexadecane), the calculated force is on the or-der of 0.1 µN, a very large force compared with optical tweezers.
4. CONCLUSION
In summary, we have experimentally and theoretically illustrated the efficiency of laser-driven localized thermocapillary stresses for mi-croactuation of water droplets. By locally heat-ing the oil-water interface, a net “pushheat-ing” force is produced, that may be used in various prac-tical microfluidic applications: Controlling the timing of drop formation, controlling the size of the drop and the path that it follows when ar-riving at a bifurcation. The generality of the process provides a practical new way for acting on individual droplets, at any location, while working inside the robust environment of the microchannel.
Several scaling laws are favorable to this forcing technique: The dominance of surface stresses over volumetric effects at small scales and the rapidity of diffusive heat transfer while thermal inertia is reduced all lead to very rapid response time. Moreover, the force due to the laser heating decreases as 1/R with the drop ra-dius R (Eq. 17) while the drag force [10] scales
as R2. Therefore the laser power necessary to
counterbalance the drag force decreases with the drop size, which also favors miniaturization.
References
[1] C.N. Baroud, J.P. Delville, F. Gallaire, and
R. Wunenburger. Thermocapillary valve
for droplet production and sorting. Physi-cal Review E, 75(4):046302, 2007.
[2] N.O. Young, J.S. Goldstein, and M.J. Block. The motion of bubbles in a verti-cal temperature gradient. J. Fluid Mech., 6:350–356, 1959.
[3] J. Chen, Z. Dagan, and C. Maldarelli. The axisymmetric thermocapillary motion of a fluid particule in a tube. J. Fluid Mech., 233:405–437, 1991.
[4] E. Lajeunesse and G.M. Homsy.
Ther-mocapillary migration of long bubbles in polygonal tubes. ii. experiments. Phys. Flu-ids, 15:308–314, 2003.
[5] K.T. Kotz, K.A. Noble, and G.W. Faris.
Optical microfluidics. Appl. Phys. Lett.,
85(13):2658–2660, 2004.
[6] B. Berge, O. Konovalov, J. Lajzerowicz, A. Renault, J.P. Rieu, M. Vallade, J.
Als-Nielsen, G. Gr¨ubel, and J.F. Legrand.
Melting of short 1-alcohol monolayers on water: Thermodynamics and x-ray scatter-ing studies. Phys. Rev. Lett., 73(12):1652– 1655, 1994.
[7] E. Sloutskin, C.D. Bain, B.M. Ocko, and
M. Deutsch. Surface freezing of chain
molecules at the liquid and liquid-air interfaces. Faraday Discuss., 129:1–14, 2005.
[8] W Boos and A Thess. Thermocapillary flow in a hele-shaw cell. J. Fluid Mech., 352:305– 320, 1997.
[9] J. Bush. The anomalous wake accompa-nying bubbles rising in a thin gap: a me-chanically forced marangoni flow. J. Fluid Mech., 352:283–303, 1997.
[10] A. Nadim, A. Borhan, and H. Haj-Hariri. Tangential stress and marangoni effects at a fluid-fluid interface in a hele-shaw cell. J. Colloid and Interface Science, 181:159–164, 1996.
[11] L. N. Trefethen. Spectral methods in mat-lab. 2000.